pounds difference in pressure between the two points, and allowance must be made for this when the height of the drop leg is fixed, in order to insure successful operation. In such a case the riser must be located near the separator, as shown in Fig. 5. The horizontal is then made a proportional length and. connected into the drop leg; then a pipe of the length necessary to connect with the boiler is laid horizontally, completing the job. Fig. 6 shows a job that is not properly designed. Two risers are connected into one horizontal, which is made twice as long as it would other wise be in order to compensate for the extra work put upon it. This is wrong, because it is quite possible for the pressure in one riser to be less than in the other, thus causing water from it to be discharged into the other separator, instead of going to the drop leg. Of course, this could be prevented by a judicious use of check-valves, but GIENCE AND INDUSTÜN, FIG. 5 such complications are undesirable, as the system should be made as simple as possible, and also because uncalled for valves do not always overcome the objections to a badly designed system. Fig. 7 shows the correct way to connect for two steam loops that are intended to discharge into one boiler, FIG. 6 as the risers and horizontals are made separate, while the two drop legs discharge into one common feed-pipe. Fig. 8 illustrates one steam loop, taking the condensation from a separator, and another taking it from a steam-heating coil, while both discharge into the same boiler. This arrangement can be made to work satisfactorily if the forgoing directions and suggestions are observed. It will be noted that in all of the illustrations here presented, the feedpipe from the steam loop enters the boiler independently of all other feed pipes. This is not an accidental plan but one that must be observed if the best results are desired, for if a pump discharges into the same. pipe, the pulsation caused by its intermittent action will interfere with the steam-loop discharge. After a system of piping that is drained by one or more steam loops has been shut down (with steam on the boilers) long enough to allow all the pipes to become filled with water, it is not reasonable to expect it to work well again until measures are taken to SCIENCE AND INDUSTRY FIG. 7 entirely remove the water of condensation from the main supply pipes, and from the risers. At other times air may collect in the top of the risers, or in the horizontals and cause trouble. A small valve should be connected at each of these points, so that it may be opened under pressure, and the air blown out. If it is desired to make this operation automatic, air valves may be used at these points, so that as soon as air begins to collect, the cooling process will cause contraction which opens them, allowing the objectionable air to escape. When steam begins to pass out, the heat causes expansion which closes the valves and prevents further loss from this source. The following calculation shows the power required to operate a steam loop under given conditions: Suppose that three boilers are in use, developing 100 horsepower each, and evaporating 9,000 pounds of water per hour into dry steam. An additional 5 per cent. of water passes out with the steam, without being evaporated, amounting to 450 pounds per hour. The difference in pressure is 20 pounds, and dividing this by the constant .41, for reasons already explained, and adding 10 per cent. to overcome fric tion, shows that the water will be raised 53.9 feet high, developing 450 x 53.9=24,255 foot-pounds per hour, which is equivalent to .0122 horsepower. The steam loop is not an economical way of developing power, but if we admit that it takes twice as much steam as the most economical pump, it is equal to only .025 horsepower, so that the whole amount is too small to be worthy of serious consideration. In some places where the difference in pressure amounts to 25 pounds or more, it may not be convenient to erect a drop leg high enough to overcome the difference in pressure, but in high buildings that are fitted with air shafts, it can easily be accomplished, and there are few shops or factories where the necessary pipes cannot be SCIENCE AND INDUSTRY. FIG. 8 run through the rooms until the required height is secured. In such cases it will not be necessary to add 25 per cent. to the height in order to overcome friction, as 10 per cent. will be sufficient. It is not always so easy to tell whether a steam loop is working or not, as in the case of a pump or an injector, but if it fails to operate the horizontal will soon become cool and thus indicate the failure. The fact that it contains no moving parts is a point in its favor, because it will last indefinitely, and its first cost is very low, compared with other apparatus used to do the same work, and it requires little or no attendance. It possesses a very great advantage over a steam trap discharging into the sewer, as it not only saves heat but puts pure water into the boilers, thus preventing the formation of scale, 80 far as the returning water is concerned. THE LARGE WATER-POWER PLANTS HE Niagara Falls Power Company, at Niagara Falls, N. Y.. shunts a portion of the water of the Niagara River around the falls by means of an inlet canal about 1,200 feet long, 250 feet wide at its mouth, and 12 feet deep; two wheel pits, each about 463 feet long, 180 feet deep, 20 feet wide, and a discharge tunnel with horse-shoe shape cross-section, 21 feet vertical diameter, 18 feet 10 inches largest horizontal diameter, and 6,890 feet long. Extension into wheel-pit No. 2 is 545 feet long. Average depth below surface, 200 feet. Average head of water, 136 feet; each turbine discharging 430 cubic feet per second; separate penstocks, 7 feet 6 inches diameter; turbines connected direct to alternators by vertical tubular steel shafts, 38 inches in diameter. The installation in power-house No. 1 consists of 10 twin turbines, each of 5,000 H. P.; two-phase, 2,200 volts, 25 cycles, 250 revolutions per minute; and power-house No. 2 will contain 11 similar units. The current here generated is transmitted to Buffalo, Tonawanda, and Lockport at 22,000 volts, three-phase, by three overhead transmission circuits, two of copper and one of aluminum conductors; each circuit about 22 miles in length. Most distant substation in Buffalo, 31.4 miles from the power-house. Thirty thousand H. P. is distributed locally, mainly on the power company's lands within two miles of the power-house, by means of 2,200volt two-phase and 11,000-volt threephase underground circuits. Fourteen miles from Bakersfield, Cal., is the plant of the Power Development Company, of San Francisco, where water from the Kern River is led through a tunnel cut in solid granite a distance of 8,484 feet, with a crosssection of 6 ft. 4 in. x 6 ft. 4 in., and having a capacity of 321 cubic feet per second. This tunnel terminates in a forebay within the mountain, and from there is conducted by a water pipe 66 inches in diameter and 600 feet long to the wheels. The total fall is 210 feet. Flow of water (mean low) is 300 cubic feet per second. The installation consists of three wheels (impulse type), one to each generator. Each wheel develops 750 H. P.; three generators of 450 K. W., 600 H. P. each; three-phase 60 cycles, 3,600 alternations, 500 volts. The current is transmitted to Bakersfield 14 miles by a six-wire pole line at 11,500 volts. From substation here pole lines extend about 20 miles. --Power. IN R. G. GRISWOLD RHEOSTAT AND GALVANOMETER IN ORDER that the strength of a current may be varied at will within certain limits, it is necessary to insert in series with the source of supply a variable resistance. The instrument performing this function is called a rheostat, and, as described herein, has a total resistance of about 47 ohms through 150 steps of .313 ohm each. It has a capacity of 6 amperes and should not be forced to carry more. Fig. 1 shows the instrument complete. The spool upon which the wire is wound is made of three circular pieces of hardwood, 6 inches in diameter by inch thick, separated by two pieces 5 inches in diameter and of an inch thick. These five pieces are glued together, heated thoroughly in an oven to expel all moisture and given three coats of shellac. Prepare three-inch strips of thin asbestos paper equal in length to the circumference of the 6-inch circles, or about 187 inches. Lay off a center line on two of them inch from the edge and running the entire length, and divide into -inch divisions. Wind these strips around the edges of the two outside disks and the unmarked one about the middle disk, cementing them in place. with a light coat of shellac. The divisions of one strip should come exactly between those on the other, as shown by the dimensions in Fig. 2. Give the strips a coat of shellac. The pins upon which the resistance coil is wound are 4-inch wire brads, driven in the two outside disks at the eighth-inch divisions, allowing them to protrude about inch. The coil is made of about 70 feet of No. 24 B. & S. softdrawn bare German-silver resistance wire, wound from one pin to the next opposite consecutive one. To facilitate the work, fasten the spool in a vise and pass the loop made at one end of the wire by soldering (b, Fig.2) over one brad. Then, while holding the wire very taut, pass it from one brad to the other. If the turns about the brads do not lie close to them, they may be made to do so by closing them with a pair of very sharp-nose pliers, easily made by grinding the ordinary kind to an edge at the end. When the last turn has been wound on, it is fastened by soldering to a small brass clip secured to the bottom disk c. Should an odd spacing of the brads bring this end to the top, fasten it there, bringing the end down through a hole in the spool to the under side, where it can be brought out in a small groove. It cannot be brought from the top to the binding screw on account of the brush arm. The protruding ends of the brads should now be sent back over the wire loops, which will serve to tighten them, and the heads cut off with a pair of cutting pliers. Then carefully file the ends down until they are just above the wire and will not interfere with the arm. Wind two or three layers of shellaced thin paper strips, inch wide, over the wires just beside the brads, and on these strips wind very tightly several turns of No. 26 B. & S. spring brass wire for binding the wire loops in place in case they should become so heated as to expand enough to slip over the brads. Solder this binding wire at several places to keep the turns together and from slipping over on to the coil. Adjust the various wires where they pass the middle strip of |